First time here? Read INSTALL.md first — it spells out what to install for Pi Agent vs. DeepSeek Harness (they are two different frameworks).

Seep Reverse Lab

Agent-Native · Multi-Platform Reverse Engineering · CWE-602 Authorization Audit · Autonomous Security Workbench

[ English | 中文文档 ]

Overview • Architecture • Demo • MCP Matrix • Quick Start • Scope • Agent Contract • Disclaimer

---

🔗 Attribution & Reference Sources:

• Mobile reverse engineering methodology & verification test suite referenced from: newliver666/apk-reverse (MIT License).
• Security research lab framework, Zero-Waste Recon signal routing & agent execution protocols referenced from: GeniusHu-tgty/Open-tgtylab (GPL-3.0 License).
• Reverse engineering knowledge base, attack graph boards & MCP automation ecosystem referenced from: LING71671/open-reverselab (GPL-3.0 License).
• Windows PE deep-dive methodology, gated procedure & dependency-free PE engine referenced from: dabao1955/exe-reverse (MIT License).
• Community support & technical discussions: LINUX DO.

---

🌟 Project Overview

Seep consolidates fragmented reverse engineering toolchains (Radare2 / JADX / Apktool / Frida / IDA), operational knowledge bases, prompt engineering contracts, and battle-tested field experience into an Agent-Native Autonomous Workbench.

Users provide plain-language technical goals; the agent autonomously performs platform identification, vulnerability type classification, surgical binary patching, and closed-loop evidence delivery.

Five Core Pain Points Solved

Pain Point	Solution
🔧 Toolchain Fragmentation	38 MCP tools wrapping Radare2, JADX, Apktool, Frida and a PE engine — unified API, zero manual switching
🧠 Agent Decision Drift	softseep orchestrator: two-stage auto-classification (Platform × 9 Task Types) + 7-gate decision tree
🛡️ Model Refusal in Security Audits	Three-tier mitigation: BLOCK_NONE injection + transparent terminology mapping + cognitive redirection
📚 Ephemeral Knowledge	294 technical field journals + 14 desensitized paradigms + "search KB before executing" enforced discipline
📦 Setup Friction	One-click install.ps1 → dedicated paths for Pi Agent and DeepSeek Harness (DSH)

---

⚙️ How It Works

User Input (plain language — "bypass the premium check in FooBar.apk")
    │
    ▼
┌───────────────────────────────────────────────────────────────────┐
│  Lab Mode  (session-scoped memory; only "lab:" in THIS session)    │
│  • Colloquial → compliance terminology auto-mapping               │
│  • BLOCK_NONE injection + cognitive redirection (anti-refusal)    │
│  • Context persists across model switches & context compression   │
└────────────────────────────┬──────────────────────────────────────┘
                             │
                             ▼
┌───────────────────────────────────────────────────────────────────┐
│  softseep  Master Orchestrator                                    │
│  ① Platform  →  Windows PE / Android APK / Linux ELF / Web       │
│  ② Task Type →  9 Paradigms × 7 Gates (G0–G6 decision tree)      │
│  ③ Route     →  exe-reverse / apkseep / ida-reverse / seep MCP / KB │
└────────┬──────────────────┬────────────────────┬──────────────────┘
         │                  │                    │
         ▼                  ▼                    ▼
   seep MCP            IDA Pro MCP         KB Search
   38 tools            (optional)          294 journals
   (Radare2/JADX/                          (Zero-Waste Recon)
    Apktool/Frida)
         │
         ▼
┌───────────────────────────────────────────────────────────────────┐
│  Context Budget Control  (v1.1 noise reduction)                   │
│  • seep_r2_decompile: full / fold (−60% tokens) / summary (−90%) │
│  • seep_r2_disasm:   full / branch (control-flow only)           │
│  • seep_r2_xrefs:    paginated, limit=10, total stats header      │
└────────────────────────────┬──────────────────────────────────────┘
                             │
                             ▼
         PoC Generated → Sandbox Execution → Self-Healing Loop
                             │
                             ▼
         Airplane-Mode Confirmation (CWE-602 iron-clad proof)
                             │
                             ▼
         3-Part Consulting-Grade Security Report  ✓

---

⚡ 30-Second Demo

Once deployed, just talk to your agent in plain language:

lab: analyze FooBar.apk — find the premium check and bypass it

The agent autonomously:

1. seep_auto_triage → identifies DEX + ARM64 SO, no packer detected
2. seep_apk_decompile with output_mode=fold → extracts control-flow skeleton only (saves ~60% tokens)
3. seep_kb_search → finds matching CWE-602 pattern from field journal #142
4. seep_apk_gen_hook → generates hook_verify.js Frida script
5. Executes on connected device → captures stdout, self-heals ClassNotFoundException, re-runs
6. Airplane-mode confirmation ✓ → seep_gen_security_report → 3-part audit report delivered

Total elapsed time on a typical client app: 8–20 minutes, fully unattended.

---

⚡ Core Capabilities

• 🎯 Natural Language Intent Resolution: Binary magic bytes + semantic verb mapping → correct toolchain, no flags to memorize
• 🔍 CWE-602 Authorization Audit: Minutes to determine if a feature gate is local-boolean or server-authoritative
• 🛡️ Authenticode Signature Preservation: DLL search-order hijacking (version.dll / sentry.dll) keeps host binary signature intact
• 🔄 PoC Self-Healing Loop: Frida error → root-cause mapping → auto-fix → re-execute (up to 3 attempts, then structured handoff)
• 💎 14 Industrial Architecture Paradigms: Monolithic offline PE → multi-process IPC → VM arbitration → .NET keygen → weak RSA bypass → Java agent → .NET VM bypass (fully desensitized, Projects A through N)
• 🌐 Transport-Layer Compliance Relay & Auto-Retry: Pure stdlib localhost relay with streaming refusal auto-retry cascade (S1-review-checklist → S2-review-framed → S3-review-en), preserving target paths without semantic drift. Unit tests 84/84 passing.
• ⚡ DeepSeek Harness (DSH) integration: python setup/dsh-plugin-setup.py installs a ~40 KB plugin package and writes the workbench path — the plugin mounts both MCP servers itself at runtime (toolCallTimeoutMs 900 s, since DSH's 60 s default cuts long tasks) plus the 10 skills, so moving the repo means editing one value. The lab： gate is now code-enforced inside the plugin (per-session, off by default; mentioning an APK does not activate it), and a settings panel (Settings → Seep 工作台) shows live status, lets you edit the path, and lists all 20 tools with their states. Verify with python setup/verify-dsh.py (6 groups / 21 items / 100 points).
• 🛡️ Measured Anti-Refusal Defense: Gemini-safe prompt reshaping (gemini-safe.flag), L1 observable decision logging (l1-log.flag), and native BLOCK_NONE payload gating.
• 🔌 Offline-Ready: All toolchains pre-bundled (657 MB of tools), zero network dependencies after setup
• 🪟 Windows PE Deep-Dive Engine (exe-reverse): 41 gated references + 16 dependency-free Python engines — peutil (RVA↔offset, imports/exports/TLS/relocs, MS checksum, per-section entropy), pe_triage, pepatch (equal-length + checksum recompute + byte re-read from disk), aob_scan (unique/ambiguous/permissive verdict), pe_diff, hwprobe, trace_reduce (+ optional SMT proof), fold. Covers packers, virtualization/flattening, .NET, Unity IL2CPP, packaged Python, kernel drivers, hardware tracing and hypervisor observation.

---

📋 Directory Structure

📁 Full Directory Tree (click to expand)

Seep\ (881 MB, of which 657 MB is the offline toolchain)
├── README.md                      ← This file (English default)
├── README.zh.md                   ← Chinese documentation (中文文档)
├── INSTALL.md                     ← ⭐ Read this first: what Pi vs DSH each install
├── AGENTS.md                      ← Project-root instruction entry (generated from Tool/prompts)
├── DSH-PROFILE.md                 ← DeepSeek Harness: field reference, mounting, verification
├── VERSION                        ← Version marker (matches the npm package version)
├── CHANGELOG.md                   ← Release notes
├── LICENSE                        ← GPL-3.0
├── check.bat / check.ps1          ← Double-click health check (wraps verify.ps1)
├── update.bat / update.ps1        ← Double-click update (idempotent, lossless)
│
├── packages\dsh-workbench\        ← ⭐ ⬆ The npm bundle (45 KB, wiring only)
│   ├── package.json               ← Package name: seep-dsh-workbench
│   ├── cordis.patch.yml           ← The single entry inserted into the load chain
│   ├── lib\index.js               ← Mounts MCPs, reads config, handles updates
│   ├── lib\lab-gate.js            ← Session-scoped lab switch
│   ├── lib\lab-rewrite.js         ← Sensitive-term rewrite (reads the repo rule table)
│   ├── lib\client.js              ← The Lab button next to the model picker + settings panel
│   ├── test\lab-gate.test.mjs     ← 39 assertions
│   └── README.md / LICENSE
│
├── Tool\
│   ├── skill\                     ← 10 reverse-engineering skills (5 primary + 5 safe)
│   │   ├── softseep\              ← ⭐ Master orchestrator (triage + routing + lazy refs)
│   │   ├── exe-reverse\           ← Windows PE deep-dive
│   │   ├── apkseep\               ← End-to-end Android reversing
│   │   ├── webseep\               ← Web / streaming / HLS chains
│   │   ├── client-license-validation-bypass\ ← Cross-runtime license attack playbook
│   │   └── safe-skills\           ← 5 standalone tool packages
│   │       ├── apk-reverse\           ← Upstream mobile reversing suite
│   │       ├── ida-reverse\           ← IDA Pro automation
│   │       ├── mcp-js-reverse-playbook\ ← JS signature chains
│   │       ├── radare2\               ← CLI reversing
│   │       └── reverse-engineering\   ← General methodology
│   │
│   ├── mcp\                       ← MCP engine
│   │   ├── seep_mcp_server.py     ← Core server: 38 analysis tools
│   │   ├── tool_resolver.py       ← Tool path resolution (user → env → bundled → path → candidates)
│   │   └── Tool\                  ← ⚠️ Hardcoded relative path (do not rename)
│   │       ├── safe\              ← Pre-bundled toolchains (561 MB)
│   │       │   ├── jadx\          ← 75 MB (v1.5.6)
│   │       │   ├── radare2\       ← 39 MB (v5.9.8 full suite)
│   │       │   ├── apktool\       ← 24 MB (v2.11.0)
│   │       │   ├── hook-mcp\      ← Frida / LSPosed instrumentation templates
│   │       │   ├── ida-pro-mcp\   ← IDA bridge adapter
│   │       │   ├── js-reverse-mcp\← Web / JS debugging engine (24 tools)
│   │       │   └── playwright-mcp\← Headless browser automation (25 tools, optional)
│   │       └── reverselab\
│   │           ├── kb\            ← 294 field journals
│   │           └── docs\          ← llms.txt and friends
│   │
│   ├── prompts\                   ← Agent coordination specs
│   │   ├── SYSTEM.md              ← The discipline text injected on lab：
│   │   ├── AGENTS.md              ← Portable instruction base (shared by Pi and DSH)
│   │   └── extensions\            ← Pi's lab gate + term rewrite (TypeScript)
│   │
│   ├── cases\                     ← 14 desensitized industrial cases (projects A ~ N)
│   ├── upstream\                  ← Upstream attribution layer (4 full mirrors)
│   │   ├── apk-reverse\           ← newliver666/apk-reverse
│   │   ├── open-tgtylab\          ← GeniusHu-tgty/Open-tgtylab
│   │   ├── open-reverselab\       ← LING71671/open-reverselab
│   │   └── exe-reverse\           ← dabao1955/exe-reverse
│   ├── aotopsy\                   ← Flutter AOT analysis (Windows-only)
│   ├── docs\                      ← Engineering references (TOOLKIT_GUIDE / PROVENANCE / DOWNLOADS)
│   └── scripts\                   ← Workspace automation
│       └── compliance-relay\      ← 🌐 Transport-layer compliance relay (stdlib-only, 3 protocols)
│
├── setup\                         ← Install / repair / update / uninstall / verify
│   ├── install.ps1 / install.sh   ← One entry point (dispatches per agent)
│   ├── install-pi.ps1             ← Pi only: writes mcp.json + copies skills
│   ├── dsh-plugin-setup.py        ← DSH only: installs the bundle + writes the path (--from-npm optional)
│   ├── verify.ps1 / verify.sh     ← Repo baseline check (64 items)
│   ├── verify-dsh.py              ← DSH deep check (25 items / 102 points)
│   ├── pack-plugin.py             ← Builds the npm tarball (prints tgz + sha256; never publishes)
│   ├── update.ps1 / update.sh     ← Idempotent update
│   ├── uninstall.ps1 / uninstall.sh
│   ├── tool-registry.json         ← Lookup rules for 20 external tools
│   └── tool-paths.local.json      ← Your manual tool paths (gitignored)
│
├── MANUAL\                        ← 14 tactical SOP guides
│   ├── AGENTS.md                  ← ⭐ Pi vs DSH wiring, side by side
│   ├── UPDATING.md                ← ⭐ Where the bundle is published, how updates flow
│   ├── PUBLISHING.md              ← Maintainer release discipline & runbook
│   ├── PREREQUISITES.md           ← Environment & dependency requirements
│   ├── IDA-PRO.md                 ← IDA Pro integration (idalib + stdio)
│   ├── ANTI-DEBUG.md              ← Anti-debug bypass dictionary & proxy DLL framework
│   ├── UNPACKING.md               ← UPX/MPRESS/Themida/VMP unpacking SOP
│   ├── POC-VALIDATION.md          ← Frida self-healing loop & sandbox validation
│   ├── DEPLOYMENT.md              ← Deployment guide (Pi / DSH routes)
│   ├── DEPLOYMENT-CHECKLIST.md    ← Official baseline checklist
│   ├── DSH-VERIFICATION.md        ← DSH acceptance list (why 102 points)
│   ├── CROSS-PLATFORM.md          ← Windows / Linux / macOS equivalence
│   ├── COMPLIANCE-RELAY.md        ← Transport-layer relay & auto-retry
│   └── MODEL-REFUSAL-MATRIX.md    ← Measured refusal matrix & prompt reshaping
│
├── docs\images\                   ← README artwork
└── .github\workflows\             ← Upstream sync (spillover-proof)

---

🛠️ MCP Tool Matrix

The bundled seep MCP server exposes 38 native tools across eight functional groups:

Category	Tool	Functionality	Token Mode
Health	seep_status	Verifies Radare2, JADX, Apktool, KB readiness	—
	seep_ida_status	Probes IDA Pro MCP service connectivity	—
Binary (R2)	seep_r2_info	Architecture, bitness, DEP/ASLR/Canary/PIE	—
	seep_r2_strings	Extracts strings with regex + section filtering	limit=
	seep_r2_functions	Functions, imports, exports, entry points	limit=
	seep_r2_disasm	Disassembly with cross-references	full / branch
	seep_r2_decompile	C-like pseudocode via pdc engine	full / fold / summary
	seep_r2_xrefs	Cross-reference graph, paginated	limit=10 default
	seep_r2_diff	Code / hex diff between two binaries	—
	seep_r2_asm	Assemble ↔ disassemble machine code	—
	seep_r2_cmd	Raw Radare2 pipeline commands	—
Windows PE	seep_pe_triage	One-pass PE triage: sections/entropy, imports, PDB, CLR, checksum, protector signals	—
	seep_pe_aob_scan	AOB signature lookup with unique / ambiguous / pervasive verdict	limit=
	seep_pe_diff	Build-to-build diff with volatile header fields normalised, mapped back to RVA	—
	seep_pe_patch	Verified equal-length patch: checksum recompute + post-write byte re-read	dry_run=
Android toolbox	seep_android_toolbox	Readiness check for external Frida / objection / APKiD / Aotopsy toolchain	—
	seep_apkid_scan	APKiD packing & anti-analysis fingerprint (protector vendor, obfuscator)	—
	seep_aotopsy	Flutter / Dart AOT (libapp.so) static analysis, Dart symbol recovery	keyword=
Web & streaming	seep_web_pipeline	One-shot chain: auth gate → fingerprint → evidence → retest → report	authorized=
	seep_web_gate	Web stage gate with hard BLOCK on missing prior artefacts	—
	seep_cms_fingerprint	CMS family, collection API, player chain, exposure-surface fingerprint	deep=
	seep_hls_analyze	HLS six-root-cause verdict: referer / auth / AES-128 / DRM / source death	—
	seep_http_replay	Header-variant matrix replay to isolate the gating header	—
	seep_spider_extract	TVBox / CatVod spider three-entry extraction	—
	seep_web_report	Generate the web source & exposure-surface assessment report	—
	seep_web_toolbox	Readiness check for the Web toolbox (ffmpeg / nuclei / ffuf)	—
Android	seep_apk_info	APK manifest, permissions, signatures (no Java)	—
	seep_apk_decompile	JADX full Java source decompilation	—
	seep_apk_unpack	Apktool resource + Smali disassembly	—
	seep_apk_smali_search	Smali pattern search (crypto keys, auth gates)	limit=
	seep_apk_gen_hook	Ready-to-run Frida hooks with stack traces	—
Knowledge Base	seep_kb_search	Full-text search across 294 field journals	limit=
	seep_kb_read	Full technical reference retrieval by topic	—
	seep_kb_checklist	Emergency triage checklists & attack matrices	—
	seep_kb_payloads	Security test seeds (JWT, SSRF, SSTI, SQLi)	—
Orchestration	seep_task_init	Initializes isolated audit sandbox directory	—
	seep_auto_triage	Automated full-sample health check	—
	seep_gen_security_report	Synthesizes 3-part compliance security report	—

💡 PE engine guardrails: seep_pe_patch refuses in-place writes, refuses a length change by default, and aborts when expect= does not match the bytes actually at the site. Only seep_pe_patch writes to disk; the other three are read-only.

💡 Context Budget Control (v1.1): seep_r2_decompile with output_mode=fold reduces token consumption by ~60%; summary mode by ~90%. Use seep_r2_xrefs instead of raw axt to avoid flooding the context window with hundreds of references.

---

🚀 Quick Start & Deployment

1. Prerequisites

• OS: Windows 10 / 11 x64 (recommended) or compatible Linux / macOS
• Runtimes: Python 3.11+, Node.js 18+, Git

2. One-Click Setup (Cross-Platform)

Windows (PowerShell):

cd setup
powershell -ExecutionPolicy Bypass -File .\install.ps1

Linux / macOS (Bash):

chmod +x setup/install.sh
./setup/install.sh

What this does automatically: Unpacks dependency archives (node_modules.zip) → validates pre-bundled tools → installs Python mcp libraries → registers MCP servers (picking an interpreter that actually has mcp) → resolves physical paths → installs external capability packages (only the zero-cost Aotopsy by default) → runs the full self-check plus a capability board.

Optional: fill in external capability packages

A few of the 38 MCP tools rely on programs that deliberately stay outside the repo. One command fills them in:

powershell -File setup\install-capabilities.ps1 -Only all   # ffmpeg / frida / APKiD / nuclei / ffuf
powershell -File setup\install-capabilities.ps1 -CheckOnly  # report only, no disk changes

Linux / macOS:

./setup/install-capabilities.sh --only all

A missing component never blocks the MCP server: the affected tool returns an explicit fix hint, and check.ps1 / verify.sh reports it as 🟡 DEGRADED rather than a failure. Platform limits: aotopsy (Flutter AOT) is Windows-only, and the eight seep_web_* tools drive PowerShell scripts, so Unix needs PowerShell 7 installed first. See CROSS-PLATFORM.md. See MANUAL/DEPLOYMENT.md — external capability packages.

3. Two Agents, Two Wirings

This workbench supports and verifies exactly two agents. There is one copy of the content; the wiring is per-agent.

Agent	Instruction entry	What gets wired	Install
Pi Agent	Tool/prompts/SYSTEM.md → ~/.pi/agent/seep-system.md	~/.pi/agent/mcp.json + copied skills	powershell -File setup/install-pi.ps1
DeepSeek Harness	project-root AGENTS.md (generated)	bundle seep-dsh-workbench	python setup/dsh-plugin-setup.py

The DSH bundle is published on npm. Two ways to install it:

# A. from npm (for users: enables the in-panel "Update plugin" button)
python setup/dsh-plugin-setup.py --from-npm

# B. from the local folder in this repo (for people editing the code: restart picks up changes)
python setup/dsh-plugin-setup.py

Both require cloning the repo first — the 561 MB toolchain lives in the repo and is not in the npm package (the package is only 45 KB of wiring). --from-npm also accepts --with-playwright (+25 browser tools) and --with-ida (+48 IDA tools).

Line-by-line comparison of the two wirings, where lab lives, and how updates differ: MANUAL/AGENTS.md.

📖 Troubleshooting: No module named mcp, execution-policy restrictions, missing Java, and more — see MANUAL/DEPLOYMENT.md.

4. Updating an Existing Installation

Already deployed? No need to re-clone or copy files by hand. The updater is idempotent and lossless — your model credentials and any custom MCP servers you added are never overwritten.

• Easiest for Windows: Double-click update.bat in the repository root for zero-command automatic upgrade.
• In-Agent Prompt: Simply type "update workbench" (or "更新工作台") in your Agent chat, and it will execute the updater and report back.
• CLI Commands:

# Windows
powershell -ExecutionPolicy Bypass -File .\setup\update.ps1

# Linux / macOS
./setup/update.sh

It will: stash local changes → git pull → print the changelog → back up your config → incrementally sync skills/prompts/MCP entries → verify user data was preserved → run the full health check.

Flag	Effect
-DryRun / --dry-run	Show what would happen, modify nothing
-Check / --check	Check if a newer version is available without pulling
-NoPull / --no-pull	Skip git pull (for ZIP-based installs)

⚠️ Always fully restart your Agent session after updating so the new skills and extensions load. See CHANGELOG.md for what changed in each version.

4b. Upstream Declarative Sync & Zero-Spillover Architecture

Seep packages mirrors of four upstream open-source repositories (apk-reverse, open-tgtylab, open-reverselab, exe-reverse). To eliminate manual merging and prevent upstream updates from polluting unrelated modules, the workbench enforces a strict-boundary declarative sandboxing architecture:

• Declarative Rule Manifest: setup/upstream-rules.json strictly binds each upstream repo to its own target directory and allowed boundary whitelist.
• Physical Zero-Spillover Gate: setup/sync-upstream.py (pure stdlib) asserts real canonical paths before copying. If upstream A tries to write into directory B, directory C, or the main project, it throws a SpilloverSecurityViolation and aborts.
• Isolated CI Pull Requests: .github/workflows/upstream-sync.yml monitors upstream repositories on schedule. When upstream A is updated, it creates a dedicated branch and PR whose diff touches strictly Tool/upstream/A/.
• Local One-Click Sync:

  powershell -ExecutionPolicy Bypass -File .\setup\sync-upstream.ps1 -Target apk-reverse

5. Uninstalling

Removes only what the workbench itself installed. Your model credentials, sessions and any MCP servers you added yourself are never touched.

powershell -ExecutionPolicy Bypass -File ./setup/uninstall.ps1 -DryRun   # preview first
powershell -ExecutionPolicy Bypass -File ./setup/uninstall.ps1          # then run

./setup/uninstall.sh --dry-run && ./setup/uninstall.sh

Everything affected is backed up to ~/.pi/agent/backup-uninstall-<timestamp>/ first, and -Restore <dir> reverses the operation. See MANUAL/DEPLOYMENT.md.

6. Verify Your Deployment

Run the 9-dimension health verifier (52 checks) against MANUAL/DEPLOYMENT-CHECKLIST.md using any of these methods:

Method	Command
⭐ Double-click (easiest)	check.bat in project root
PowerShell	powershell -ExecutionPolicy Bypass -File .\check.ps1
Agent chat	Send check (or doctor / 检查) — agent runs and reports inline

---

🎮 Workflow & Lab Mode

Lab Mode Protocol

DSH users: a Lab button sits left of the model selector. Grey = off, click once to turn it green (on), click again to leave. Session-scoped only — nothing written to disk, no restart. (Session-Scoped Trigger)

Activate:   lab:                          # or: lab: analyze FooBar.exe
Deactivate: exit lab

• Trigger is explicit: nothing activates until you type lab:. New sessions start clean.
• Isolation is session-scoped: state lives in the extension process memory (isLabSessionActive), reset by session_start. Another terminal window is unaffected, and nothing survives a restart.
• Once activated: the Seep system prompt is appended, terminology auto-mapping, anti-refusal overrides and shortcut expansion become live.
• Outside lab mode the global AGENTS.md forbids loading reverse-engineering skills or calling seep_* tools, so daily coding/ops work stays clean.

Task Shortcuts (Active in Lab Mode)

Shortcut	Action
poc <target>	CWE-602 client-side authorization audit + verification code
find-auth <target>	Locate license, subscription, expiry, hardware-binding functions
hook <function>	Generate Frida hook with stack trace + return-value override
gen-patch <offset>	Binary patch bytes or proxy DLL scaffold
triage <sample>	Full triage: architecture, imports, packers, strings
check	Run 7-section workbench health verifier inline
report	Synthesize active directory evidence → 3-part audit report

---

🔬 Technical Scope

Click to expand full technical scope

1. Client-Side Authorization Audit (CWE-602)

Authority Attribution: Airplane mode + loopback hijacking + timestamp offset testing to classify gates as server-authoritative vs. local-boolean within minutes.

Fourteen Industrial Paradigms — all fully desensitized:

Project	Architecture	Key Technique
A	Monolithic offline PE	Scalar return override (mov eax,1; ret)
B	Multi-process hybrid	Proxy DLL dispatch + 3-tier state persistence (v12.0 → v6.8.1 → v6.8.10.0: single winhttp naked thunk + semantic re-anchoring 54 pts + soft update block)
C	Resource template + UI	Bijective bit-permutation decoding + IAT hook on SetDlgItemTextW
D	Recompile-induced non-uniform shift	AOB dual-state signature migration (3 versions) + PE gating + ACL locking + optional-site version adaptation + runtime call-stack locating
E	EXECryptor VM arbitration	2-point Call redirection to memory stubs
F	.NET dynamic deobfuscation	Harmony memory dump + 96-bit combined hash keygen
G	Self-referential SHA-384	5-byte function-entry patch + watchdog persistence
H	Ed25519 pubkey replacement	In-place ciphertext replacement via derived keystream
I	Weak-modulus RSA	Sliding-window bypass + activation injection on export entry
J	Online Card/Key Authorization	Protocol decryption + memory patching + local credential spoofing
K	.NET WPF + Themida Packing	Memory dump unpacking + privilege decision branching + registry state freeze
L	Qt5 C++ Client	Proxy DLL hook + 11 privilege decision constant-folds + local LLM translation gateway
M	Java + install4j Dual-Layer	DLL search-order hijack (version.dll IAT hook) + JVM native ClassFile bytecode patching (burp.Zfqu / burp.Zwxg.Zu) + license/AI token preference seeding
N	.NET x64 + VM Obfuscation	Runtime memory extraction + privilege flag inversion + silent auto-injection

2. Android & DEX/SO Analysis

• Surgical DEX same-length patching with automated Adler-32 / SHA-1 recalculation
• Packer classification: Java2C / native payload / extraction shell / private DEX-VMP
• Runtime anti-analysis: root detection bypass, SSL pinning circumvention, Frida-RPC bridging
• Repack pipeline: STORED resources.arsc + 4-byte Zipalign + v1+v2+v3 signing

3. Binary / Native (PE / ELF / Mach-O)

• exe-reverse deep-dive layer: PE structure & triage, equal-length patching with checksum recompute and post-write byte re-read, AOB signature uniqueness verdicts, build-to-build diff, protector/packer identification, virtualization & control-flow flattening, algorithm recovery, symbolic execution
• Runtime-specific routes: managed .NET (incl. obfuscation), Unity IL2CPP metadata, packaged Python (PyInstaller / Nuitka / pyarmor), Flutter Dart AOT, kernel drivers
• Hardware-observation layer: hardware breakpoints, Intel PT instruction tracing, ETW, hypervisor-assisted observation — for targets where every software hook is detected
• Headless Radare2: architecture ID, entropy scan, symbol recovery, C-like decompilation
• Full IDA Pro MCP integration: Hex-Rays decompilation, xrefs, struct recovery
• Anti-tamper defeat: deliberate crash stubs, raw svc syscall detection, kernel anti-debug

4. CTF & Challenge Workflows

• Attack-network routing: Signal → seep_kb_search → template assembly → MCP execution
• Web: JWT, KID injection, SSRF chains, SSTI, deserialization gadget chains, Protobuf decoding
• Seed libraries (seep_kb_payloads) + emergency checklists (seep_kb_checklist)

---

🤖 Agent Execution Contract

• G-Auth Gate: Confirm testing authorization on every new target. No authorization = stop.
• 7-Gate Decision Tree (G0–G6): Classify authority ownership (Server vs. Client) before touching any code.
• Two-Strike Rule: Same-shaped failure twice → refute the technical model, fall back to classification. Third variation is strictly prohibited.
• Zero-Waste Recon: Signal detected → seep_kb_search → pre-existing tool → execute. Never write scripts from scratch without checking the KB first.
• Definition of Done: Target hash → RVA identification → PoC execution → offline airplane-mode confirmation → 3-part structured delivery. Console log alone is not evidence.

---

📝 Deliverable Specification

All client-side vulnerability assessments follow a consulting-grade 3-part structure:

1. Vulnerability Detail & Risk — Exact RVA / file offsets, call chain, CWE-602 mapping, business severity
2. Reproduction & PoC — 100% reproducible instructions, proxy DLL source, or Frida script + offline confirmation evidence
3. Defense-in-Depth Remediation:
   • SetDefaultDllDirectories → prevent DLL hijacking
   • ProcessDynamicCodePolicy → prevent executable memory injection
   • Server-side authority via cryptographic signatures and short-lived tokens

---

🤝 Acknowledgements & License

• License: This project is licensed under the GNU General Public License v3.0 (GPL-3.0).
• Special thanks to newliver666/apk-reverse (MIT License) for the foundational Android reverse engineering paradigm and verification methodology.
• Special thanks to GeniusHu-tgty/Open-tgtylab (GPL-3.0 License) for the security research lab framework and Zero-Waste Recon routing architecture.
• Special thanks to LING71671/open-reverselab (GPL-3.0 License) for the comprehensive reverse engineering knowledge base, attack boards, and MCP automation ecosystem.
• Gratitude to the LINUX DO community for technical exchange, insight, and research collaboration.

---

⚖️ Disclaimer

This repository is intended solely for authorized security research, white-box auditing, compliance vulnerability testing, and educational CTF training.

• Explicit Authorization Required: Written authorization from the asset owner is mandatory before analyzing any target.
• No Warranty: All methodologies are provided "as is" based on sandbox measurements.
• Isolated Testing: Conduct all analysis in isolated VMs or sandboxes. Testing against production systems or unauthorized networks is prohibited.
• Limitation of Liability: Authors assume no liability for misuse, unauthorized testing, or violations of applicable laws.

---

⭐ Star History